Anthropic Says Newer Models Are "Less Likely to Begin an Attack," Drawing Ridicule
AIFlow_ML · x · 2026-10-23
StockMKTNewz relays Anthropic's statement that its newer models are less likely to "begin an attack," which AIFlowML mocks: "so they are stupid like 4.8." The vague phrasing of the safety claim drew ridicule and debate in the AI community about how such statements are worded and measured.
More from Fun
- Award-winning Spanish novel found 99% AI-generated despite contest's AI ban — TuhinChakr · 2026-10-23
- Gemini's 'Argon' model refuses to analyze an Excel report, claims it's text-only — Human-League3524 · 2026-10-23
- X feeds filling up with AI cut-off videos that end abruptly mid-watch — rickasaurus · 2026-10-23
- Valve engineer used ChatGPT to find Deadlock's new matchmaking algorithm — Aryvyo · 2026-10-23
- The academic paradox: the more famous your name, the less time for the work — prof_kamilov · 2026-10-23
- Everyone questions ranking methodology publicly and celebrates rising three spots privately — prof_kamilov · 2026-10-23